Low-frequency (<0.01 Hz) oscillations in the surf zone longshore current with wavelengths too small (<300 m) to be surface gravity waves were observed during the 1986 SUPERDUCK experiment at Duck, North Carolina. The observations suggest that these oscillations are dynamically linked to the mean longshore current in the surf zone, leading Bowen and Holman to propose that the observed oscillations are manifestations of a shear instability in the longshore current. In this paper, field data from both the SUPERDUCK experiment (a barred beach) and the l980 NSTS experiment, at Leadbetter Beach, Santa Barbara, California (a plane beach), are used to compare quantitatively the model of Bowen and Holman (which is extended to include the effects of dissipation due to bottom friction) with observation. Observed frequency-cyclic wavenumber (f-K) spectra (constructed from alongshore arrays of velocity measurements made in about 1.5–2 m of water in the trough of the bar at SUPERDUCK, and in about 1 m at NSTS) are compared with theoretical predictions. Quantitative agreement is found between observation and theory at SUPERDUCK, where these motions dominate the observed f-K spectra, theoretical growth rates of the temporal instability tend to be large, and the mean longshore current varied between 0.35 and 1.0 m s−1. This comparison supports the shear instability hypothesis. Results from Leadbetter Beach (where the mean longshore current was always less than 0.5 m s−1) are less conclusive because of scatter in the observed f-K spectra; however, observation and theory agree on the f-K quadrant and on the general area within it. Finally, stability properties at the two beaches are compared. It is shown that longshore current profiles over a plane and a barred beach may give rise to different stability properties, suggesting that shear instabilities may be a more common feature on barred beaches.